Background: Interventions to improve inpatient sleep rely on the ability to objectively quantify sleep; however existing methods of measurement (wrist actigraphy and patient survey) are resource intensive and impose a burden on patients. To overcome these barriers, we developed “sleep opportunity” (SLOP), a surrogate metric for sleep derived solely from the electronic health record (EHR). We have previously demonstrated the clinical utility of this metric as it correlates with delirium incidence and can be modified through reduction of unnecessary nighttime interruptions. However, this metric had yet to be compared to previously validated metrics as a true proxy for sleep. In this study, we compare SLOP with wrist actigraphy and patient survey, the current standard tools for the measurement of inpatient sleep.
Methods: For each patient night in the hospital, SLOP is defined as the longest span of uninterrupted time between 10pm and 6am. It is determined by time-stamped interventions recorded in the EHR that correspond with a member of the healthcare team physically entering the patient room. Interruptions are excluded if they are not within control of the primary clinician, such as patient-driven medication requests (one-time or “as needed” doses).Between September 2018 and November 2019, we enrolled 66 adult patients admitted to the Hospital Medicine service at an 800-bed academic urban teaching hospital. Patients receiving sedative medication, experiencing delirium (NuDESC > 0), or who were non-English speaking were excluded from the study. For each patient, a single night of sleep between 10pm and 6am was measured using using wrist actigraphy and patient survey (Karolinska Sleep Quality Index). Collected sleep data included total sleep duration, efficiency, and fragmentation. The data from wrist actigraphy and patient survey were then correlated with SLOP using nonparametric analysis.
Results: Preliminary statistical analysis included 13 nights of wrist actigraphy and 11 nights of patient survey data, which were compared with SLOP. The sleep parameters of total duration and efficiency correlated positively with SLOP, while fragmentation correlated negatively.
Conclusions: In this study, we have begun the validation of a new metric for sleep quality that relies solely on the EHR. Early results suggest that SLOP is a true proxy for sleep in the hospital. Analysis of the full dataset will strengthen the validation of the EHR-derived metric.Compared with other methods to measure inpatient sleep, SLOP requires minimal resources and eliminates patient burden. Additionally, we previously demonstrated that SLOP correlates with clinical outcomes and can be modified through intervention. Given the metric’s ease of use and clinical utility, it has the potential to transform and expand the study of sleep in the hospital.